---
library_name: scikit-learn
tags:
- regression
- tabular-regression
- fuel-consumption
- fuelcast
---

# FuelCast Regression Model

This model predicts `Consumer_Total_MomentaryFuel` from ship, propulsion, and
weather variables from the FuelCast dataset.

## Training

- Dataset: `krohnedigital/FuelCast`
- Config: `cps_poseidon`
- Model: `random_forest`
- Rows used: `30000`
- Target: `Consumer_Total_MomentaryFuel`

## Metrics

| Metric | Value |
| --- | ---: |
| R2 | 0.9957 |
| RMSE | 0.0295 |
| MAE | 0.0194 |

## Files

- `fuelcast_model.joblib`: complete scikit-learn pipeline
- `metrics.json`: evaluation metrics
- `input_schema.json`: expected raw input columns
- `sample_input.json`: example input row
- `feature_importances.csv`: feature importance when available

## Usage

```python
from huggingface_hub import hf_hub_download
import joblib
import pandas as pd

model_path = hf_hub_download(
    repo_id="username/fuelcast-model",
    filename="fuelcast_model.joblib",
)
model = joblib.load(model_path)

sample = pd.DataFrame([{"Ship_SpeedOverGround": 12.4}])
prediction = model.predict(sample)
```